/**
* Returns a function that filters a document map based on the given include and exclude rules.
*
* @see #filter(Map, String[], String[]) for details
*/
public static Function<Map<String, ?>, Map<String, Object>> filter(
String[] includes, String[] excludes) {
CharacterRunAutomaton matchAllAutomaton = new CharacterRunAutomaton(Automata.makeAnyString());
CharacterRunAutomaton include;
if (includes == null || includes.length == 0) {
include = matchAllAutomaton;
} else {
Automaton includeA = Regex.simpleMatchToAutomaton(includes);
includeA = makeMatchDotsInFieldNames(includeA);
include = new CharacterRunAutomaton(includeA);
}
Automaton excludeA;
if (excludes == null || excludes.length == 0) {
excludeA = Automata.makeEmpty();
} else {
excludeA = Regex.simpleMatchToAutomaton(excludes);
excludeA = makeMatchDotsInFieldNames(excludeA);
}
CharacterRunAutomaton exclude = new CharacterRunAutomaton(excludeA);
// NOTE: We cannot use Operations.minus because of the special case that
// we want all sub properties to match as soon as an object matches
return (map) -> filter(map, include, 0, exclude, 0, matchAllAutomaton);
}
/**
* Build the {@link CharacterRunAutomaton} that represents the reindex-from-remote whitelist and
* make sure that it doesn't whitelist the world.
*/
static CharacterRunAutomaton buildRemoteWhitelist(List<String> whitelist) {
if (whitelist.isEmpty()) {
return new CharacterRunAutomaton(Automata.makeEmpty());
}
Automaton automaton = Regex.simpleMatchToAutomaton(whitelist.toArray(Strings.EMPTY_ARRAY));
automaton =
MinimizationOperations.minimize(automaton, Operations.DEFAULT_MAX_DETERMINIZED_STATES);
if (Operations.isTotal(automaton)) {
throw new IllegalArgumentException(
"Refusing to start because whitelist "
+ whitelist
+ " accepts all addresses. "
+ "This would allow users to reindex-from-remote any URL they like effectively having Elasticsearch make HTTP GETs "
+ "for them.");
}
return new CharacterRunAutomaton(automaton);
}
